Degron tagging to label membrane-wrapped objects and probe membrane topology

Visualization of specific organelles in cells over background fluorescence can be challenging, especially when reporters localize to multiple places. Instead of trying to identify proteins enriched in specific membrane-wrapped objects, we used a selective degradation approach to remove reporters from the cytoplasm. Using the ZF1 degron in C. elegans embryos, we demonstrate specific labelling of organelles, including extracellular vesicles, as well as individual neighbouring membranes. We show that degron tagging facilitates long-term tracking of released cell debris and cell corpses, even during their uptake and phagolysosomal degradation. We also demonstrate how degron tagging can probe the topology of the nuclear envelope and plasma membrane during cell division, giving insight into protein and organelle dynamics. As degron tags are used in bacteria, yeast, plants, and animals, this approach will enable the specific labelling and tracking of proteins, vesicles, organelles, cell fragments, and cells in many model systems.


Introduction
Membranes form barriers that separate the cell from its environment and separate diverse subcellular compartments so that they can carry out distinct functions within cells 1 . As membrane bilayers are less than 10 nm in diameter, light microscopy techniques often struggle to distinguish neighbouring membranes 2 . Membranes are also highly dynamic, undergoing fusion and fission events during diverse budding processes, including endocytosis, vesicular trafficking, and ectocytosis 3 . Using light microscopy, it is difficult to tell when membranes have completed budding and have a closed topology. Therefore, techniques that allow specific labelling are required to visualize membrane dynamics and probe membrane topology in living cells.
Fluorescent imaging techniques are only as specific as the reporters used to visualize objects. Reporters binding specific phosphatidylinositol species are popular for studying membrane dynamics 4 , but these lipids are often present on multiple objects, making it hard to distinguish or track individual objects. One example is during phagocytosis, when cellular debris or cell corpses are engulfed by the plasma membrane 5 . Both the corpse and engulfing cell plasma membranes contain the same phosphatidylinositol species, making it challenging to distinguish these membranes in living cells. Electron microscopy and super-resolution light microscopy can visualize the few tens of nm that separate the phagosome membrane from the corpse membrane, but these techniques rely on fixation, which makes it challenging to study dynamics 6 . Another example is extracellular vesicles released from cells. Vesicles that bud from the plasma membrane by ectocytosis contain the same protein and lipid species as the membrane they originate from 7 , which makes it hard to distinguish released vesicles from the plasma membrane of neighbouring cells, as well as to follow the release and uptake of these vesicles in vivo. Thus, new approaches are needed to specifically label these objects.
To overcome these challenges, we repurposed selective degradation to visualize specific membrane objects. Degron-mediated degradation decreases target protein levels and is an alternative to RNA interference or genetic knockouts 8 . Degron motifs recruit ubiquitin ligases to polyubiquitinate target proteins, resulting in the proteasomal degradation of cytosolic proteins or the lysosomal degradation of transmembrane proteins 9 . Rather than using degron tagging as a loss-of-function technique, we used the degradation of degron-tagged reporters in the cytosol to specifically label certain cells, cell fragments, organelles, and vesicles.
Here, we used the zinc finger 1 (ZF1) degron from the C. elegans PIE-1 protein to drive the degradation of fluorescent reporters in developing C. elegans embryos. The ZF1 degron is a small 36 amino acid motif recognized by the SOCS-box protein ZIF-1, which binds to the elongin C subunit of a ubiquitin ligase complex 10 . ZIF-1 is expressed in sequential sets of differentiating somatic cells 11 , resulting in a stereotyped pattern of degradation 10 . Fusing ZF1 to a target protein results in rapid degradation within 30 to 45 min of ZIF-1 expression 12 . Here, we show that ZF1-tagged reporters protected by membranes are no longer accessible to ubiquitination and degradation. This results in background-free labelling of specific cells, organelles, and vesicles. This improvement in the signal-to-noise ratio enabled the visualization of extracellular vesicles in vivo, the long-term tracking of individual phagosomes, as well as distinguishing a corpse plasma membrane from the phagosome membrane in vivo. In addition, ZF1-tagging allowed us to measure the timing of nuclear envelope breakdown and abscission during cell division. The use of degron-tagged reporters thus provides a convenient method for investigating in vivo dynamics from the level of proteins to cells.

Results
To determine whether degron-tagged reporters would be useful for cell biological approaches, we first tested whether an endogenous degradation system was capable of degrading abundant reporter proteins and examined whether the increased proteasomal load had negative effects on cells. To show that degron tags can rapidly degrade membrane reporters, we tagged the PH domain of rat PLC1∂1 with the ZF1 degron from C. elegans PIE-1 and expressed it in worm embryos. Similar to an mCherry-tagged PH reporter (Fig. 1A), the cytosolic mCh::PH::ZF1 reporter initially localized to the plasma membrane (Fig. 1D). Thus, the degron tag did not disrupt the normal localization of the reporter.  11,13 . These polar bodies maintain mCh::PH::ZF1 fluorescence (arrowheads in Fig. 1). Thus, degron tagging was able to rapidly degrade a bright, exogenous reporter in cells where the reporter could be ubiquitinated.
As ZIF-1 has a number of known targets, whose proteasomal degradation is important for embryonic development 10 , we tested whether the expression of ZF1-tagged reporters disrupted embryonic development. Stable transgenic strains expressing various ZF1-tagged reporters were fertile and had viable progeny that did not show a delay in cell cycle timing (Fig.   S1). This suggests that degron reporters are well tolerated and do not disrupt embryonic development.

Degron reporters detect released extracellular vesicles
In addition to labelling the plasma membrane, mCh::PH and mCh::PH::ZF1 labelled endocytic vesicles within cells (arrows in Fig. 1). Some of these vesicles maintained their fluorescence in the mCh::PH::ZF1 strain (video 1). As mCh::PH::ZF1 localized on the cytosolic face of vesicles would be accessible for ubiquitination and proteasomal degradation, the persistence of the degron reporter suggests that it is protected from proteasomal degradation by intervening membranes. We hypothesized that the PH::ZF1 reporter persists in extracellular vesicles or other cell debris that are taken up by the cell by endocytosis. As extracellular vesicles are derived from cytoplasm and cellular membranes, it is difficult to distinguish their contents from the cell using conventional reporters 14 .
We tested whether the degron-tagged PH reporter could be used to specifically label and track microvesicles released in vivo. Microvesicles are 90-500 nm vesicles that arise from plasma membrane budding, also called ectocytosis 7 . In wild type C. elegans embryos, microvesicles are difficult to detect using light microscopy due to their low abundance and proximity to the plasma membrane. Microvesicle budding is normally inhibited by the TAT-5 lipid flippase, resulting in increased microvesicle release when tat-5 is knocked down 15 .
Therefore, we compared the localization of mCh::PH and mCh::PH::ZF1 after tat-5 RNAi treatment. In mCh::PH embryos, microvesicle release is visible as thickened membrane labelling between cells (Fig. 1H-J) in comparison to control embryos ( Fig. 1A-C). However, small patches of microvesicles are difficult to detect over the background of the plasma membrane fluorescence. In contrast, released microvesicles are clearly visible using the mCh::PH::ZF1 reporter ( Fig. 1L-M) 3 , due to proteasomal degradation of the plasma membrane label (Fig. 1N). Released extracellular vesicles are also visible moving in the fluid around cells in the space between the tat-5 RNAi-treated embryo and the eggshell (Video 1). Thus, using a general plasma membrane reporter with a degron tag, it is possible to observe microvesicles and their movement in vivo.
We also tested whether it was possible to label extracellular vesicles by degron-tagging transmembrane proteins. In contrast to the proteasomal degradation of cytosolic proteins, ubiquitination of transmembrane proteins leads to endocytosis and lysosomal degradation 9 . The syntaxin SYX-4 is a single-pass transmembrane protein that localizes to the plasma membrane and endocytic vesicles ( Fig. S2A-C) 16 . Degron-tagged GFP::ZF1::SYX-4 shows a similar localization before the onset of ZIF-1 expression (Fig. S2D), after which GFP::ZF1::SYX-4 is lost from the plasma membrane and accumulates in intracellular vesicles (Fig. S2E). These vesicles eventually disappear from ZIF-1-expressing cells (Fig. S2F), consistent with ubiquitindriven endocytosis and lysosomal degradation (Fig. S2I). To test whether lysosomal degradation of a transmembrane protein can also be used to label extracellular vesicles, we treated the GFP::ZF1::SYX-4 reporter strain with tat-5 RNAi to induce microvesicle release.

Degron reporters enable tracking of phagocytosed cargo
To test whether degron tags facilitate the long-term tracking of labelled objects, we chose to observe the two polar bodies in which ZF1 degradation does not occur ( Fig. 1D-F). As polar bodies are dying cells, they have a nucleus and can be tagged with chromosome reporters like histone H2B, in addition to the PH domain 13 . Both polar bodies are initially found on the anterior surface of the embryo ( Fig. 2A, D). The first polar body is trapped in the eggshell 17 , while the second polar body (2PB) is phagocytosed by one of the anterior cells 13 . Because mCh::H2B labels all nuclei in the embryo, it can be challenging to track the 2PB phagosome among the many dividing nuclei ( Fig. 2B-C, Video 2). In contrast, the degron-tagged ZF1::mCh::H2B reporter disappears sequentially from somatic nuclei, leaving the two polar bodies as the only fluorescent objects on the anterior half of the embryo (Fig 2E, Video 2). This confirms that the intervening cell and phagosome membranes protect ZF1-tagged proteins from proteasomal degradation (Fig. 2G). Improving the signal-to-noise ratio with degron tagging also improves automated tracking of the 2PB by removing overlapping traces (Fig. 2C, F).
Tubulation of the 2PB phagosome into smaller vesicles can also be easily followed with degrontagged reporters (video 1, left) 13 . Thus, degron tagging facilitates tracking and reveals organelle dynamics by removing background labelling.

Degron reporters reveal membrane topology and dynamics
In addition to facilitating tracking of labelled phagosome cargo, degron tagging also allows the specific labelling of a single membrane or surface of a membrane. Using light microscopy, it is difficult to distinguish the signal of the corpse plasma membrane from the engulfing phagosome membrane, due to their close proximity 18,19 . Using degron reporters, this can be achieved in vivo by degrading the reporter localized to the phagosome membrane, leaving only the corpse membrane labelled (Fig. 3B). For example, the 2PB membrane initially appears as a hollow sphere in an unlabelled cell using the degron-tagged mCh::PH::ZF1 reporter (Fig. 3A). However, the plasma membrane of the polar body corpse must be disrupted in order to degrade or recycle corpse contents within the safety of the phagosome membrane 20 .
Corpse membrane breakdown can be visualized using the mCh::PH::ZF1 reporter, which is seen dispersing throughout the phagosome lumen ( Fig. 3A') 13 . Thus, by removing fluorescent reporters from membrane surfaces facing the cytosol, degron-tagging enables examination of specific membranes and their dynamics.
In addition to plasma membrane and endosomal proteins, degron tagging can also be used to assess the topology of proteins associated with the nucleus. We degron-tagged the nuclear lamin LMN-1 to examine the dynamics of the nuclear cortex during cell division 21 . Similar to proteins protected by a vesicle membrane, LMN-1 is largely protected from ZF1-mediated proteasomal degradation by the nuclear envelope ( Fig. 4I) and appears as a hollow sphere during interphase (Fig. 4A). During mitosis, the nuclear envelope breaks down for chromosome segregation 22 , which also allows ZIF-1 to target degron-tagged nuclear proteins ( Fig. 4J). During division of ZIF-1-expressing anterior cells, an mKate2::ZF1::LMN-1 reporter is degraded from anterior nuclei (Fig. 4C, video 3), disappearing by the time nuclear envelope breakdown is visible by DIC (Fig. 4D, H). Thus, degron-tagging is a probe for membrane topology that can be used to investigate the dynamics of nuclear envelope breakdown.
Degron tagging can also reveal information on protein import/export across membranes. Nuclear lamins are translated in the cytosol and imported into the nucleus, where they integrate into the nuclear matrix 23

Degron reporters reveal membrane topology during abscission
In order to understand how degrons in restricted spaces can be ubiquitinated and degraded, we applied degron tagging to the process of abscission. During cell division, the actomyosin furrow closes around the spindle midbody to form a narrow intercellular bridge 24 .
Both sides of the bridge are cleaved during abscission to release a ~1 µm extracellular vesicle called the midbody remnant, which is later phagocytosed (Fig. 5H) 25 . The intercellular bridge no longer permits diffusion between cells ~4 minutes after furrow ingression 24 , but the first cut for abscission does not occur until ~10 minutes after furrow ingression 26 . Therefore, we asked when the actomyosin accumulated in the intercellular bridge was accessible for proteasomal degradation. We degron-tagged non-muscle myosin (NMY-2) and measured the fluorescence intensity of NMY-2::GFP::ZF1 in the bridge between the anterior daughter cells (Fig. 5A) 25 .
Degradation of cytoplasmic NMY-2::GFP::ZF1 was first visible 8 ± 1 minutes after furrow ingression. NMY-2::GFP::ZF1 in the bridge showed a small but significant decline for the next 2 minutes (Fig. 5G), suggesting that NMY-2 is normally able to diffuse out of the bridge up to 10 minutes after furrow ingression. Subsequently, NMY-2::GFP::ZF1 in the bridge was protected from proteasomal degradation (Fig. 5B, G), suggesting that either a diffusion barrier had formed or abscission had occurred. Thus, we could confirm the timing of abscission estimated from electron microscopy data from fixed embryos using degron reporters and light microscopy on living embryos.
To test whether degron tagging was able to detect novel phenotypes in abscission mutants, we depleted proteins implicated in abscission, including the ESCRT-I subunit TSG-101 and the septin UNC-59. At the onset of ZF1-mediated degradation, NMY-2::GFP::ZF1 labelled intercellular bridges in tsg-101 or unc-59 mutants normally (Fig. 5C, E). In contrast to control embryos where NMY-2::GFP::ZF1 fluorescence persisted through midbody release and phagocytosis (Fig. 5B), NMY-2::GFP::ZF1 fluorescence intensity continued to drop significantly in the bridge of both tsg-101 and unc-59 mutants (Fig. 5D, F-G), and this drop was dependent on ZIF-1 expression 25 . Phagocytosis of the midbody remnant was also delayed in both tsg-101 and unc-59 mutants (Fig. 5D, F) 24,25 , consistent with a delay in abscission. These findings demonstrate that NMY-2 is able to diffuse out of the bridge and be degraded when abscission is delayed (Fig. 5I). As no defect was detected for tsg-101 knockdown using a dextran diffusion assay 24 , our ability to detect a defect in abscission demonstrates how highly sensitive degron tagging assays can be.

Discussion
In summary, degron-mediated degradation is more than a loss-of-function technique; it is a powerful tool to study dynamics from the level of proteins to organelles to cells. By removing the cytoplasmic fluorescence, degron tags improve the visibility of extracellular, luminal, or nuclear reporters and enable long-term tracking. Degron-tagged reporters also reveal insights with a standard epifluorescence microscope thought to be limited to superresolution or electron microscopy on fixed samples. Our studies have focused on objects that are protected by membrane bilayers, but this approach should work for any structure resistant to ubiquitination or diffusion. Thus, degron tagging is an important addition to the cell biologist's toolbox.
As degron-tags are widely used in cell extracts, cell culture and in vivo, our approach can help to visualize objects in many experimental systems. We chose ZIF-1-mediated degradation of ZF1 degrons for its simplicity, because it only required the expression of a degron-tagged reporter in C. elegans embryos. A variation on this technique expressing a GFP nanobody fused to the ubiquitin ligase adapter ZIF-1 enables spatial control of degradation of any GFP-tagged protein in C. elegans at any stage 27 . Ubiquitin-mediated degradation can also be regulated by small molecule drugs, temperature, or light 8 , offering many modalities to control the timing of protein degradation. Auxin-inducible degradation (AID) is a powerful option used in mammalian cultured cells and assorted animal models 28,29 . AID requires the coexpression of the degron-tagged construct with a ubiquitin ligase subunit, plus the addition of a plant hormone. This combinatorial approach enables both temporal and spatial control of 8 degradation. Thus, degron tagging can readily be applied to probe cell biology in many model systems.
We used degron tags to label and track objects for which conventional reporters are insufficient. For example, extracellular vesicles are typically detected by the tetraspanin proteins on their surface, but tetraspanin content is heterogeneous among extracellular vesicle subpopulations 30 . By degron-tagging membrane-associated or transmembrane proteins, we were able to specifically label extracellular vesicles in vivo, which has proved to be a valuable tool to screen for new regulators of extracellular vesicle budding 3 . Although our PH::ZF1 reporter is likely to favour plasma membrane-derived extracellular vesicles (microvesicles), it is possible to target endosome-derived extracellular vesicles (exosomes) by degron-tagging proteins associated with the endosome surface or by degron-tagging transmembrane proteins, given that ubiquitination drives endocytosis of transmembrane proteins (Fig. S2I). Degrontagging reporters found at both the plasma membrane and endosome surface, such as actinbinding proteins, should label both microvesicles and exosomes 31 . Thus, degron-tagged reporters are a novel approach to label and track extracellular vesicles in vivo and beyond, which will enable further studies into their function.
Wide-field light microscopy is normally limited to detecting objects that are >200 nm away from each other 32 . At this resolution, neighbouring membranes cannot be distinguished.
In addition to specifically labelling extracellular vesicles next to the plasma membrane, we showed that degron-tagged reporters could distinguish the cargo corpse membrane from the engulfing phagosome membrane. This enabled the visualization of corpse membrane dynamics during phagolysosomal clearance using wide-field microscopy 13 . These reporters enable the precise staging of phagosomes for approaches such as correlative light and electron microscopy (CLEM) 33 , which can be used to determine the ultrastructure of membrane breakdown during phagolysosomal clearance. Therefore, degron-tagging is a useful tool to reveal novel insights into organelle dynamics in addition to the long-term tracking of specific cells in vivo.
The limited resolution of wide-field microscopy also prevents the observation of membrane fusion and fission events, including during cell division. Using a degron-tagged reporter, we were able to measure early changes in the topology of the nuclear envelope as well as in the restricted space of the intercellular bridge, which was previously only possible using electron microscopy 26 . We were also able to detect defects in abscission after knocking down an ESCRT subunit 25 , which were not visible using cytosolic diffusion assays 24 . Degron tagging is therefore a sensitive tool to study the topology of membranes while they undergo fusion or fission.
Degron tagging also revealed protein dynamics and topology. Knocking down septins or TSG-101 led to distinct rates of degron-mediated degradation of the myosin reporter, which could indicate that the intercellular bridge is open to differing degrees in these mutants, 9 resulting in different rates of diffusion out of the bridge. We also found that a pool of the degron-tagged lamin reporter underwent degradation during interphase. As ubiquitination can lead to nuclear export through nuclear pore complexes 34 , this may indicate that the degron is ubiquitinated in the nucleus during interphase, which leads to export of the reporter into the cytosol, where it is then accessible for proteasomal degradation. Alternatively, this may be due to the activity of the nuclear proteasome 35 . As degradation of a reporter depends on its location within the cell, the degron approach can potentially be applied to determining the topology of transmembrane proteins (type I/II) or to distinguish cytosolic from luminal proteins. Thus, degron tags can be used to analyse protein import/export into organelles and determine the topology of proteins associated with membrane-bound organelles. In summary, degron tagging improves the signal-to-noise ratio, reveals super-resolution insights on a standard microscope, and provides insights into localization and dynamics from the level of cells to proteins. We thank Katrin Heinze, Sonja Lorenz, and Anna Liess for critical reading of this manuscript.

Author contributions
AMW conceived and designed the study. KBB, GF, LI, and AMW performed and interpreted experiments. KBB, GF, and AMW wrote the manuscript.

Competing Interests
The authors declare no competing interests.

Worm strains and maintenance
Caenorhabditis elegans strains were maintained according to standard protocol at room temperature 36 . For a list of strains used in this study, see Table S1. unc-59 loss-of-function mutant embryos were generated by feeding unc-59 RNAi to the WEH132 strain bearing a hypomorphic unc-59 mutation 25 .

Worm transformation
FT205, WEH251, and WEH399 were made by biolistic transformation using a Bio-Rad PDS-1000, as described 37 . The DP38 strain was bombarded with MP322 (gift of Michael Glotzer 15 ) to generate FT205. WEH251 was generated by co-bombardment of pGF04 13

Light Microscopy
Embryos were dissected from gravid adults and mounted in M9 buffer on an agarose pad on a glass slide. For FT205 and FT368, imaging was performed using a Zeiss AxioImager, 40X 1.3 NA objective, an Axiocam MRM camera, and AxioVision software, as described previously 15 .
For WEH02, WEH51, WEH132, and WEH142, Z-stacks were acquired sequentially for GFP and mCherry every 20 seconds, as described 25 . For strains WEH260 and WEH296, Z-stacks were acquired for mCherry every minute at room temperature, as described 13 . For WEH251, Zstacks were acquired for mCherry every 20 seconds using a Leica DM5500 wide-field fluorescence microscope with a HC PL APO 40X 1.3 NA oil objective lens supplemented with a Leica DFC365 FX CCD camera controlled by LAS AF software. For WEH399, Z-stacks were acquired for mKate2 and DIC every 30 seconds. Time-lapse series were analysed using Imaris (Bitplane).

Cell cycle timing
To compare the speed of development between control and ZF1-tagged strains, the time from the 6-to the 12-and to the 24-cell stages (one and two subsequent cell cycles in the AB lineage) were calculated from time-lapse series. The cell cycle time was measured from the ingression of any furrow in the two ABx cells to the ingression of any furrow in the four ABxx or eight ABxxx cells. To estimate percent changes, we compared the timing of ABx division to ABxx division for reporter strains with and without the ZF1 degron.
Tracking H2B-labeled nuclei were tracked over time using the surface function of Imaris with thresholding to define objects.

Quantification of corpse membrane breakdown
Corpse membrane topology was measured using a box scan across the 2PB phagosome in the WEH260 strain. A box with 3-pixel thickness was drawn through the middle of the phagosome using Fiji (NIH) and the mean profile intensity was measured.

Fluorescence intensity measurements
Mean fluorescence intensity of the LMN-1 reporter (Fig. 4K) was measured in a circle with an area of 0.8 µm 2 using ImageJ (NIH) in ABp and EMS nuclei. Fluorescence intensity was measured from the first time point that EMS nucleus reformed in a round shape after P1 division until the LMN-1 marker in ABp nucleus was not distinguishable from the cytoplasm.
Fluorescence intensity of the EMS nucleus was measured as an internal control. Data are reported as the ratio of the fluorescence intensity of the ABp nucleus to that of the EMS nucleus.
A sigmoidal decay curve with tilted baseline was fitted to the data using OriginPro (OriginLab) and the slopes of the initial mild fluorescence loss during interphase and the sharp drop during mitosis were calculated.
Mean fluorescence intensity of the NMY-2::GFP::ZF1 reporter (Fig. 5G) was measured in a circle with an area of 0.5 µm 2 using ImageJ (NIH), as described previously 25 . Midbody fluorescence was measured from contractile ring closure until the end of the time lapse series or until the midbody was not distinguishable from the cytoplasm. Fluorescence intensity of the first polar body was measured as an internal control. An exponential decay curve was fit to the polar body data using OriginPro and used to correct for fluorescence loss due to photobleaching.
Embryos were excluded if the P0 and AB midbodies were too close to each other or if the polar body data did not fit an exponential decay function. NMY-2 data are reported as the ratio of the fluorescence intensity of the midbody to the expected value of the polar body after cytoplasmic background subtraction. Timing of the onset of degradation of NMY-2::GFP::ZF1 in the cytoplasm was judged by eye by comparing the relative brightness of ABp and EMS using Imaris.

Image Processing
For clarity, images were rotated, colorized and the intensity was adjusted using Adobe Photoshop. All images show a single optical section (Z), except for Fig. 2 and Fig. S2. In Fig.   2, six Zs with 1.2 µm steps were maximum projected using Leica LAS X software. In Fig. S2, images where maximum projected to span a region of 2.5 µm using Fiji (NIH). Fig. 2C, 2F, and videos were rotated, colorized, and the intensity was adjusted using Imaris. Several Zs were maximum projected in Imaris for videos. For video 3, the brightness was adjusted in each frame using Photoshop to compensate for photobleaching.

Statistical evaluation
Student's one-tailed t-test was used to test statistical significance. When necessary, Bonferroni correction was used to adjust for multiple comparisons.           Table   Table S1: Strains used in this study. This study